speaker-route

name	speaker-route
description	Route Jarvis audio responses to a Sonos speaker. Mic stays on Discord/phone — only the output is redirected. Supports per-message routing and a persistent speaker mode that stays active until turned off.
category	voice
runtimes	["claude"]
pii_safe	true
tier	JARVIS
triggers	["say that on the speaker","play that on the speaker","send that to the speaker","route to speaker","speaker mode on","speaker mode off","switch to kitchen speaker","switch to bedroom speaker","follow me to the kitchen","follow me to the bedroom","play through the speaker","say that downstairs","say that upstairs","stop speaker mode","turn off speaker mode"]

speaker-route — Route Jarvis Audio to Sonos

Discord/phone stays the input. Jarvis's spoken response comes out of whichever Sonos speaker you choose instead of (or in addition to) the phone.

Examples

"Say that on the speaker." "Switch to bedroom speaker." "Follow me to the kitchen — speaker mode on." "Speaker mode off." "Play that downstairs."

Speaker Targets

User says	Target	sonos-say arg
"kitchen", "downstairs"	Kitchen Sonos	down
"bedroom", "upstairs", "bathroom"	Bedroom Sonos	up
"everywhere", "all speakers", "whole house"	Both	all
(no location, mode on)	Default (down)	down

Modes

Per-message routing

User says "say that on the speaker" or "send that to the bedroom" — route only the current response, then return to normal.

After generating your response text:

1. Call sonos-say <target> "<response text>"
2. Keep the Discord text reply as normal
3. No persistent state change

Speaker mode (persistent)

User says "speaker mode on" or "switch to [speaker]" — all subsequent Jarvis responses also play on that speaker until turned off.

1. Write the target to /tmp/jarvis-speaker-mode (e.g. down, up, or all)
2. Confirm: "Speaker mode on — routing audio to [speaker]. Say 'speaker mode off' to stop."
3. On every subsequent response: read /tmp/jarvis-speaker-mode, call sonos-say with the response text

To check if speaker mode is active:

cat /tmp/jarvis-speaker-mode 2>/dev/null

Empty or missing = mode off.

Turn off speaker mode

User says "speaker mode off" or "stop speaker mode":

rm -f /tmp/jarvis-speaker-mode

Confirm: "Speaker mode off — audio back to Discord only."

Response Routing

When routing to speaker, the response text should be:

• Natural spoken language (no markdown, no bullet points, no emoji)
• Concise — trim any lengthy formatted sections to a spoken summary
• The same Jarvis voice: warm, direct

Keep the full text response in Discord as usual. The speaker gets a clean spoken version.

Script

sonos-say <up|down|all> "response text"

Persistent State File

/tmp/jarvis-speaker-mode   — contains: up | down | all

Missing or empty = speaker mode off.

Always Check on Startup

At the start of every response, check if speaker mode is active:

SPEAKER_TARGET=$(cat /tmp/jarvis-speaker-mode 2>/dev/null)

If non-empty, append a sonos-say call after generating your response.